用时间深度学习对儿科质瘤的纵向风险预测
Divyanshu Tak1,2, Biniam A Garomsa1,2, Anna Zapaishchykova1,2
1Artificial Intelligence in Medicine (AIM) Program, Mass General Brigham, Harvard Medical School, Boston.
NEJM AI
|June 19, 2025
概括
这项研究引入了一个时间深度学习模型,可以显著改善使用连续MRI扫描来预测儿科质瘤复发的情况. 人工智能方法提高了风险评估,有可能优化患者监测和对脑瘤的护理.
科学领域:
- 在瘤学中使用人工智能
- 医学成像分析 医学成像分析
- 儿科神经瘤学 儿科神经瘤学
背景情况:
- 儿科质瘤的复发呈现异质的模式,具有挑战性的预测当前的临床和基因组标记.
- 由于不可预测的复发,儿童质瘤患者的频繁,长期MRI监测是标准的.
- 有限的数据可用性和现有的机器学习方法阻碍了个性化复发预测的进展.
研究的目的:
- 开发和验证一种深度学习方法,以使用纵向MRI数据更好地预测儿科质瘤复发.
- 加强对正在接受监测的儿科质瘤患者的个性化风险评估.
- 探索时间学习对其他癌症和慢性疾病的适应性.
主要方法:
- 开发了一种自主监督的时间深度学习模型,用于纵向医学成像.
- 模型编码了连续MRI扫描,训练在时间顺序分类 (借口任务).
- 用715名患者 (3994次扫描) 的历史监测扫描进行小儿质瘤1年复发预测的微调模型.
主要成果:
- 与传统方法相比,时间学习提高了复发预测性能 (F1评分) 高达58.5%.
- 在低度和高度儿科质瘤中观察到性能增长,AUC在75%至89%之间.
- 随着更多的历史扫描,预测准确性得到改善,在3到6次扫描之间停滞不前.
结论:
- 时间深度学习为儿科脑瘤监测和决策支持提供高性能纵向分析.
- 这种方法在跟踪和预测其他癌症和慢性疾病的风险方面具有广泛应用的潜力.
- 人工智能模型在神经瘤学中促进了更精确,更个性化的患者管理.
相关概念视频
02:57Targeted Cancer Therapies
The targeted cancer therapies, also known as “molecular targeted therapies,” take advantage of the molecular and genetic differences between the cancer cells and the normal cells. It needs a thorough understanding of the cancer cells to develop drugs that can target specific molecular aspects that drive the growth, progression, and spread of cancer cells without affecting the growth and survival of other normal cells in the body.
There are several types of targeted therapies against specific...
There are several types of targeted therapies against specific...
01:21Cancer Survival Analysis
Cancer survival analysis focuses on quantifying and interpreting the time from a key starting point, such as diagnosis or the initiation of treatment, to a specific endpoint, such as remission or death. This analysis provides critical insights into treatment effectiveness and factors that influence patient outcomes, helping to shape clinical decisions and guide prognostic evaluations. A cornerstone of oncology research, survival analysis tackles the challenges of skewed, non-normally...